Patterns of Chasmogamy and Cleistogamy, a Mixed-Mating Strategy in an Endangered Perennial

Research Article

Patterns of chasmogamy and cleistogamy, a mixed-mating strategy in an endangered perennial

Stephanie M. Koontz1*, Carl W. Weekley1, Sarah J. Haller Crate2 and Eric S. Menges1 1The Plant Ecology Program, Archbold Biological Station, Lake Placid, FL 33862, USA 2Longleaf Program Coordinator, North Carolina Forest Service, North Carolina Department of Agriculture and Consumer Services, Raleigh, NC 27699, USA

Received: 16 December 2016 Editorial decision: 14 October 2017 Accepted: 30 October 2017 Published: 7 November 2017 Associate Editor: Joseph H. Williams, Jr. Citation: Koontz SM, Weekley CW, Haller Crate SJ, Menges ES. 2017. Experimental evidence for benefit of self discrimination in roots of a clonal plant. AoB PLANTS 9: plx059; doi: 10.1093/aobpla/plx059

Abstract. Cleistogamy (CL) in angiosperms historically has been understudied; however, its co-occurrence with chasmogamy (CH) across many plant species suggests a fitness advantage to maintaining this mixed-mating strategy. Maintenance of mixed-mating has been attributed to reproductive assurance, resource allocation or genetic trade-offs. Our goals were to explore patterns of CH and CL, quantify reproductive contributions measured by fruit production and determine how CL is maintained in the endangered perennial Polygala lewtonii. This species exhibits CH and both above-ground cleistogamy (CL-AG) and below-ground cleistogamy (CL-BG). In monthly censuses from 2008 to 2012, we documented flowering patterns by counting CH flowering stems, CL-AG fruits and CL-BG rhizomes per plant. Monitoring of buds on CH flowering stems in 2004 provided an estimate of CH fruits per plant. Plant excavations in 2005 of CL-BG rhizomes provided an estimate of CL-BG fruits per plant. Floral morphs were temporally separated with CH flowers observed from January to May and CL flowers from June to February. Overall, 17.5 % of plants flowered; most plants expressed CH first in spring months (63.4 %) and the rest initiated CL-AG in fall months. Reproductive output was dominated by CH (median 26 fruits) compared to combined CL (median 3.5 fruits). Annual reproductive effort of CL-AG was positively correlated with plant age while CH had no rela- tion. Our research shows CH as the dominant form of reproductive effort with most individuals expressing CH and through greater reproductive contributions. CL appears limited by plant size or resources based on the positive relationship with plant age. CL dependency on resource availability is common in other species found in dry or low- quality habitats; however, CL contributions in this species are comparatively low. This raises more questions related to energy requirements of both floral morphs, how this affects the production of viable progeny and why CL persists.

Keywords: Amphicarpy; chasmogamy; cleistogamy; flower dimorphism; resource availability; spatial and temporal variation.

Introduction progeny while preserving locally adapted alleles. This mixed-mating was once thought to be evolutionar- Mixed-mating in plants can provide a unique fitness ily unstable; however, multiple reproductive strategies advantage through the production of genetically diverse occur frequently in vascular plants, with 42 % of species

*Corresponding author’s e-mail address: [email protected]

© The Author(s) 2017. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 1 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

examined exhibiting some form of mixed reproduction more effective at purging deleterious alleles (Lande and (Goodwille et al. 2005). In a separate review, cleistogamy Schmske 1985; Charlesworth and Charlesworth 1987) (CL; closed, self-pollinating flowers) was documented in while genetically more diverse outcrossed progeny are 693 species across 50 families and 77 % of these also maintained through heterosis (Lu 2002; Oakley et al. had chasmogaous, presumably outcrossed flowers 2007). (dimorphic CL; Culley and Klooster 2007). The frequency Here, we document the pattern of mixed-mating in the in which mixed-mating strategies have occurred sug- rare Polygala lewtonii (Polygalaceae), a federally endan- gests there is strong selection pressure for producing gered perennial herb (USFWS 1999; Coile and Garland mixed progeny. 2003) found on only two ancient sand dune ridges Strategies for mixed-mating systems can occur with (Mount Dora and Lake Wales Ridges) in central Florida. the production of open-pollinated chasmogamous Its primary habitat is sandhill, dominated by long- flowers and permanently closed self-pollinating cleis- leaf pine-wiregrass assemblages on xeric yellow sands, togamous flowers (Lord 1981; Schoen and Lloyd 1984; an ecosystem shaped by frequent fire (1–10 years; Myers Culley and Klooster 2007), with both self-compatible 1990; Menges 1999) and seasonal fluctuations in rain- and incompatible individuals (Stone 2002) or with indi- fall (dry winters and wet summers) and temperature vidual flowers forming either outcrossed or selfed fruits (summer temperatures > 30 °C; Menges 1999). Polygala (Schoen and Brown 1991). Chasmogamous flowers, lewtonii adults are killed by fire but seedlings recruit when cross-pollinated, produce genetically diverse pro- post-fire from a persistent soil seed bank (Weekley and geny, thus maintaining genetic diversity, while cleistog- Menges 2012). In the absence of fire, populations de- amous progeny possess only maternal information and cline and may disappear above-ground. preserve locally adapted genes (Schoen and Lloyd 1984; Polygala lewtonii is one of three species within the Waller 1984; Mitchell-Olds and Waller 1985; Schmitt family Polygalaceae exhibiting CL (Lord 1981; Culley et al. 1985; Winn and Moriuchi 2009). Chasmogamy and Klooster 2007). Both P. polygama and P. pauciflora (CH) usually relies on pollinator availability for pollen exhibit amphicarpy and observational studies suggest transfer, although self-pollination is also possible in their mixed-mating systems are maintained through re- some species. Chasmogamous flowers typically are en- source allocation (Shaw 1901). CL in P. lewtonii was first ergetically more expensive to produce and have lower briefly described by James (1957) as the species’ ability seed set compared to cleistogamous flowers (Schemske to set seed in both open and closed flowers. Small dark 1978; Waller 1979; Schoen and Lloyd 1984; Mitchell-Olds purple to pink chasmogamous flowers are clustered on and Waller 1985). CL increases a populations’ suscepti- terminal racemes. Chasmogamous flowers rely on in- bility to genetic drift and inbreeding depression if dele- sect pollinators and delayed selfing is rare (Weekley and terious alleles cannot be purged (Lloyd 1979; Lande and Brothers 2006). Aerial cleistogamous flowers are incon- Schemske 1985). These fitness trade-offs are factors in spicuous, green to pale pink and solitary in the lower leaf maintaining a mixed-mating strategy. axils. Subterranean CL occurs on rhizomes extending There are several hypotheses explaining natural from the base of the plant. A recent study examining the selection leading to the maintenance of mixed-mating spatial genetics of P. lewtonii suggested most recruit- strategies (Goodwillie et al. 2005; Oakley et al. 2007). ment is from cleistogamous seeds (Swift et al. 2016). Reproductive assurance describes selfing as a backup For the remainder of this study we refer to chasmogamy mechanism when pollen is limiting or stochastic events as CH, above-ground cleistogamy as CL-AG and below- occur (Le Corff 1993; Masuda and Yahara 1994; Culley ground cleistogamy as CL-BG. 2002). Here, production of cleistogamous flowers is Understanding reproductive patterns and limita- dependent on the relative success of CH and floral tions of P. lewtonii can provide needed insight to its morphs are separated temporally or spatially to ensure reproductive ecology and better inform conservation progeny success (Berg and Redbo-Torstensson 1998). efforts. The goal of this study was to characterize CH Another hypothesis is that allocation of resources to dif- and CL in this rare Polygala. Our objectives were to (i) ferent floral morphs optimizes the use of available energy describe flowering trends and frequencies of all three reserves (Schemske 1978; Schoen and Lloyd 1984). With floral morphs (CH, CL-AG and CL-BG) since they have not resource allocation, production of both floral morphs is been previously described for P. lewtonii, (ii) quantify the independent of each other but one or both are correlated reproductive output of the three floral morphs and (iii) with a resource, typically size (Waller 1980; Steets and explore the selective pressures associated with main- Ashman 2004) or pollinator availability (Culley 2002). taining a mixed-mating strategy. Mixed floral morphs may also be stable by a genetic bal- As previously discussed, there are several hypoth- ance between selfing and cross-pollination. Selfing is eses supporting the maintenance of mixed-mating.

2 AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

We predict that if CL occurs as a reproductive assur- plant at each monthly census. CL-BG rhizomes were ance strategy, there would be a negative correlation counted by carefully excavating around the base of with chasmogamous fruit production. However, if CL is multi-stemmed (>1 vegetative stem) plants and count- maintained by partitioning of available energy reserves ing any extending rhizomes. The presence of rhizomes (resource allocation hypothesis), we expect cleistog- suggested below-ground reproductive activity but entire amous fruit production to correlate with plant size or rhizomes were not excavated to count flowers and fruits age but to have no direct association with chasmoga- due to the destructive nature of such excavations. Mean mous fruit production. We also explore pessimistic (ini- fruit production per rhizome of CL-BG was estimated tiating CL first) and optimistic strategies (initiating CH using a separate data set (see below). first) of initiating one floral morph over another, as first described for annual grasses (Zeide 1978; Cheplick and Estimating fruit production Quinn 1982) but also observed in perennial grass spe- To compare reproductive effort of these three floral cies (Campbell et al. 1983). These two strategies balance morphs, counts of mature fruits per plant were needed. between producing genetically unique offspring (CH; op- During 2008–2012 monthly census, the total number timistic) and producing any offspring at all (CL; pessim- of mature CL-AG fruits were counted but not for CH or istic) in stochastic conditions. CL-BG. To obtain median fruit production per plant for CH and CL-BG, we used two separate data sets collected from the same site in 2004–2005. Methods In March 2004, 20 plants were randomly selected to We followed individuals of P. lewtonii at the Lake Wales monitor CH flower and fruit development as part of an- Ridge National Wildlife Refuge (LWRNWR) Carter Creek other study (Weekley and Brothers 2006). On each plant, located in south-central Florida (Fig. 1). This refuge is three random bud-bearing CH stems were marked using predominantly xeric sandhill and is managed with pre- colour-coded thread for identification and the number scribed fire, most recently occurring in 2001, 2007 and of buds per stem counted. Plants were monitored for 2009; the 2009 prescribed fire did not affect our study 59 days as buds developed. Fruits were mature if abscis- area. Data on individual plants and counts of CL-AG ma- sion occurred with a slight touch or within 22 days after ture fruits were collected from 2008 to 2012. Specific initiation. Using these data, we calculated a median counts of mature fruits on CH stems and CL-BG rhizomes number of mature fruits produced per CH stem. were collected in 2004 and 2005, respectively, on sep- In April 2005, 40 plants of various sizes were selected arate individuals. to excavate for below-ground rhizomes. All plants were carefully excavated to reveal the full length of all rhi- Data collection zomes and to quantify all reproductive structures (buds, Plants were followed monthly from germination to flower, fruits and capsules). Not all rhizomes had below- senescence from March 2008 through December 2012 ground structures and we assumed all reproductive (excluding July 2008) for a total of 57 censuses. Indivi structures matured. We used these data to quantify the dual plants were marked with pin flags and unique num- median number of mature fruits per rhizome. bered aluminium tags. Seedling survival in this species is low with >50 % of a Statistical analysis cohort dying within their first year (Weekley and Menges We used descriptive statistics to compare median age of 2012) making it difficult to maintain an adequate sample reproductive maturity and the frequency of each floral size. To combat this problem, we added new recruits dur- expression, and a chi-squared test to determine the ing monthly censuses as needed throughout the study. probability of initiating a floral morph. Therefore, plants followed in this study are not from the same cohort but were all followed from germination to Estimating reproductive output. CL-AG flowers are senescence or until the study’s termination. small, inconspicuous and mature rapidly; therefore, At each monthly census, we recorded survival, counts may underestimate true CL-AG production. CH counted vegetative stems and quantified reproductive flowering stems and CL-BG rhizomes often persist from activity of the three flower morphs. Vegetative stems month to month meaning some stems may have been were counted on all individuals until CH flowering stems counted twice, overestimating reproductive output for formed, after which only active CH flowering stems these two morphs. To reduce over estimates, the peak were counted. The number of mature CH fruits were number of CH stems and CL-BG rhizomes observed dur- not counted but estimated using a separate data set ing each reproductively active season was used as the (see below). CL-AG mature fruits were counted for each maximum number of reproductive stems/rhizomes for

AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 3 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

Figure 1. Map showing location of Polygala lewtonii populations at the Lake Wales Ridge National Wildlife Refuge Carter Creek along the Lake Wales Ridge in Highlands County, Florida.

that individual for that season. We acknowledge repro- annual CH and CL-AG fruit production and determined ductive output varies by individual and year. Our esti- how fruit production for both floral morphs varied with mates of CH and CL-BG fruit production are based on plant age. In this study, we used plant age as a proxy data collected from different plants in a different year for plant size based on marked individuals in perman- however, more exact counts were not feasible due to ent plots. A separate demographic data set showed time constraints and the destructive nature of excavat- 14 years of size measurements taken annually in March ing CL-BG rhizomes. determined plant age and size are correlated in P. lew- We calculated the number of mature fruits produced tonii (N = 1287, Pearson’s r = 0.32, P < 0.0001; Weekley per plant based on the annual peak number of CH stems and Menges 2012). Both models included individual as observed during monthly censuses. The same was done for a random effect to account for repeated measures on CL-BG using the annual peak number of rhizomes observed. individual plants. Regression analyses were run on sig- This gave us an estimate of peak annual fruit production for nificant fixed factors to determine the direction and CH and CL-BG per plant to compare with CL-AG. strength of any significant relationships identified in mixed models. All fruit counts were natural log trans- Hypotheses maintaining mixed-mating. Linear mixed formed to fit normality assumptions. All analyses were models examined the relationship between peak done in SPSS version 22.0 (IBM Corp. 2013).

4 AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

Results median, we estimated CH fruit production to be a median of 26 fruits per plant ranging from 13 to 182 Our study captured data on 234 seedlings of P. lewtonii fruits (mean = 45.6 ± 40.9 SD). from germination to senescence with 17.5 % (N = 41) Fifteen of the 40 plants had 1–10 rhizomes (me- surviving to reproductive maturity. Reproductive ma- dian = 3) with a median of 1 reproductive structure per turity was defined as flowering CH or CL-AG at least stem (range 0–14; mean = 2.8 ± 3.3 SD). Assuming all once; some plants expressed only a single floral morph. below-ground reproductive structures matured, we cal- Seedling recruitment occurred year-round with peak culated a median of 1.5 fruits per plant with a range of germination in spring months; therefore, our sample 1–6 fruits (mean = 2.3 ± 1.7 SD). size was biased with most seedlings germinating from Plants with CL-AG produced a range of 1–42 mature February through April also surviving to reproduction fruits with a median of 2 per plant (mean 4.4 ± 7.7 SD) (Table 1). compared to 1.5 CL-BG fruits and 26 CH fruits. Even the combined output of CL (3.5 mature fruits) was less than Flowering trends and frequencies CH reproductive effort. Among observed plants, 24.4 % produced only CH flowers, 9.7 % produced only CL-AG flowers, 41.5 % flowered Hypotheses maintaining mixed-mating both CH and CL-AG and 24.4 % flowered all three floral CH and CL flower and fruit production showed temporal morphs (Table 2). Median age for initiating CH flowers separation with little overlap (Fig. 2), making P. lewto- was 23 months compared to 19 months for CL-AG and nii reproductively active all year. CH flowering stems 29.5 months for CL-BG. There was a marginally sig- were found from January to May. CL production fol- nificant difference in the probability of initiating CH or lowed shortly after with CL-AG fruits found from June to CL-AG first (χ2 = 2.951, df = 1, P = 0.086); almost two- January and CL-BG rhizomes from July to February. thirds of plants initiated CH first (63.4 % vs. 36.6 % CL-AG Linear mixed models showed there was a marginally first;Table 2 ). No plants in this study initiated CL-BG first. significant relationship between CH and CL-AG fruit production (F = 3.95, P = 0.051) with significant variation Estimating reproductive output 1, 62.3 in production of both floral morphs between individuals Mature CH fruits per stem ranged from 0 to 24 with (Wald Z = 3.024, P = 0.002). Regression analysis showed a median of 13 fruits (mean = 12 ± 5 SD). Using this a weak positive but significant relationship between 2 CH and CL-AG fruit production (r = 0.23, F1, 81 = 23.968, P < 0.001; Fig. 3). Table 1. Recruitment months of 234 Polygala lewtonii seedlings, Peak seasonal fruit production varied significantly by their survival to reproduction and percent of plants initiating CH floral morph (F1, 117.7 = 107.52, P < 0.001) and the inter- first. Otherwise, plants initiated above-ground CL first. No plants action of morph with age (F = 6.09, P = 0.015) but not produced below-ground CL first. 1, 115.9 age alone (F1, 135.0 = 0.81, P = 0.368). Fruit production also Month # recruits % survived to % flower varied by individual (Wald Z = 3.19, P = 0.001). Breaking reproduction CH first down the interaction by floral morph shows no significant effect of age on CH fruit production (F = 2.24, January 4 0 % – 1, 72.2 P = 0.139) but a significant effect on CL-AG fruit pro- February 50 14.0 % 71.4 % duction (F1, 67.7 = 11.41, P = 0.001). Regression analyses March 69 29.0 % 65.0 % showed a weak but significant positive relationship 2 April 29 13.8 % 25.0 % with plant age and CL-AG fruit production (r = 0.16, F = 13.9, P < 0.001) and no relationship with CH fruit May 21 19.0 % 100 % 1, 66.3 production (r2 = 0.04, F = 3.98, P = 0.050; Fig. 4). June 0 – – 1, 74 July 0 – – Discussion August 7 29.6 % 50.0 % Mixed-mating systems have evolved independently September 13 0 % – many times with several hypotheses addressing how October 6 16.7 % 100 % these systems are maintained. Our results show a mar- November 0 – – ginal probability for initiating chasmogamous flowers December 35 8.6 % 33.3 % before above-ground cleistogamous flowers, greater Total 234 17.5 % 63.4 % chasmogamous reproductive effort, temporal separation between CH and CL flowering periods and a

AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 5 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

Table 2. Expression of CH, above-ground and below-ground cleistogamy (CL) in 41 reproductive individuals of Polygala lewtonii followed from

germination through senescence. Frequency shows the percentage of plants that expressed either a single or multiple flower morphs (N1 = number of plants producing each combination of floral morphs). Initial floral morph shows the percentage of plants that initiated CH or CL

first (N2 = number of plants that first initiated a specific floral morph first). Median age and range in months for when each floral morph was

first observed regardless of initial floral morph (N3 = number of plants that expressed each floral morph).

Floral morph Frequency (N1) Initial floral morph N( 2) Median age in months; range (N3)

Chasmogamy (CH) 24.4 % (10) 63.4 % (26) 23; 7–45 (37) Above-ground cleistogamy (CL-AG) 9.7 % (4) 36.6 % (15) 19; 11–41 (31) Below-ground cleistogamy (CL-BG) 0 % (0) 0 % (0) 29.5; 16–44 (10) CH and CL-AG 41.5 % (17) – – CH, CL-AG and CL-BG 24.4 % (10) – –

Figure 2. Monthly CH and CL flowering/fruiting patterns of Polygala lewtonii from 2009 to 2012. Plants were monitored monthly and recorded as reproductively active with chasmogamously (CH) flowering stems, above-ground cleistogamous (CL-AG) fruits or rhizomes on which below- ground cleistogamous (CL-BG) fruits are found.

positive relationship between plant age and AG-CL to CH in one-third of plants, with some individuals never fruit production. Temporal separation of floral morphs producing chasmogamous flowers. Thus, CL in P. lewto- suggests CH and CL are favoured under different sea- nii appears to be resource dependent while CH may be sonal conditions and have different resource limitations restricted by another resource such as pollinator avail- (Schoen and Lloyd 1984). We found no evidence for CL ability (Culley 2002). acting as reproductive assurance for failed chasmoga- In many species, CL is expressed as a response to mous production. Instead, CL was positively correlated resource availability. Plant size has been shown to be with CH indicating no trade-off in resource allocation to a limiting factor for reproductive effort (Jasieniuk and either floral morph. Finally, chasmogamous fruit produc- Lechowicz 1987; Diaz and MacNair 1998; Munguía- tion was estimated to be over seven times that of cleis- Rosas et al. 2015) with some studies demonstrating that togamous fruits (above- and below-ground combined). manipulations of above-ground vegetation can signifi- CL has been shown to occur prior to CH in response cantly reduce CL alone (Diaz and MacNair 1998) or both to resource stress (Cheplick and Quinn 1982; Campbell floral morphs (Munguía-Rosas et al. 2015). Resource et al. 1983) or after adequate resources and growth have limitations that negatively impact CL may be induced been obtained in both annual and perennial grasses by environmental stresses such as soil moisture or soil (Zeide 1978; Cheplick and Quinn 1982). Our data show fertility gradients (Schoen and Lloyd 1984; Bell and CL was positively related to plant age and occurred prior Quinn 1987; Albert et al. 2011). However, in a review by

6 AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

Campbell et al. (1983), two-thirds of perennial non-rud- Winn and Moriuchi 2009). Cleistogamous seeds in P. eral grass species expressed CL in response to stochastic lewtonii are larger than chasmogamous seeds (C. W. environmental conditions. Weekley, Archbold Biological Station, pers. comm.), al- In P. lewtonii, chasmogamous fruit production was though successful seed set and progeny fitness of both over seven times greater than that of above- and below- floral morphs were not explored in this study. Limited ground CL combined, a pattern that does not fit with CL suggests poor environmental conditions outside the other studies of CL. Most species with mixed-mating ideal growing season but under more favourable con- have higher cleistogamous seed production or produce ditions all modes of reproduction are more successful larger seeds, and several studies found that cleistog- (Jones et al. 2015). High chasmogamous reproductive amous progeny out-performed chasmogamous pro- effort was observed in the current study and high chas- geny across several developmental stages (Schemske mogamous fruit maturation was found by Weekley and 1978; Clay 1983; Waller 1984; Sun 1999; Culley 2002; Brothers (2006) even with low insect visitation. However, autogamy is rare in P. lewtonii, based on low chasmogamous fruit maturation in a pollinator exclusion experi- ment (Weekley and Brothers 2006). At this time, we have no explanation for the discrepancy in observed chasmogamous and cleistogamous fruit maturation. The mixed-mating system of P. lewtonii is separated temporally, with selfing by CL positively associated with resource availability. It is still peculiar that chasmogamous reproductive effort exceeds cleistogamous efforts the reverse of what is seen in most other species. A recent population genetic study found that most individuals surviving to adulthood are progeny from cleistogamous seeds (Swift et al. 2016), suggesting a higher fitness for cleistogamous progeny. This also raises the question of why so much effort is being put Figure 3. Linear regression model of maximum annual mature into CH if few outcrossed progeny are represented in fruit production for chasmogamous (CH) and above-ground cleis- the next generation of reproductive individuals. More togamous (CL-AG) flower morphs ofPolygala lewtonii (N = 63, some research is needed to understand the apparent failure points overlap). The regression shows a significant positive rela- of CH to produce viable offspring, even as more effort tionship in mature fruit production between the two flower morphs is allocated to producing these seeds. Additionally, we (r2 = 0.23, P < 0.001). need a better understanding of how environmental gradients affect resource allocation for reproduction. These two key topics would add valuable knowledge to the reproductive biology of P. lewtonii and aid in highlighting conservation concerns (such as limited pollinator availability) for this species.

Conclusions The occurrence of mixed-mating should be evolution- arily unstable, but has been documented in many vascular plants (Goodwillie et al. 2005; Culley and Klooster 2007) including in the rare P. lewtonii. We found a pattern of strong seasonal separation between CH and CL flower production, a positive correlation between mature cleistogamous fruit production and plant age, and initiation Figure 4. Linear regression model of plant age and maximum an- of CH in two-thirds of plants prior to CL. These patterns nual fruit production of chasmogamous (CH) and above-ground have been observed in other species with mixed-mating cleistogamous (CL-AG) flower morphs. There was a significant relationship of CL-AG fruit production (r2 = 0.16, P < 0.001) with age and in low-quality habitats with variable rainfall (Campbell a marginal relationship of CH fruit production with age (r2 = 0.04, et al. 1983; Jones et al. 2015) and are linked to resource P = 0.050). requirements. Temporal separation in floral morphs

AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 7 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

allows P. lewtonii to be reproductively active year-round Campbell CS, Quinn JA, Cheplick JP, Bell TJ. 1983. Cleistogamy in in a stochastic environment dominated by dynamic rain- grasses. Annual Review of Ecology and Systematics 14:411–441. fall events and fluctuating temperatures, but raises con- Charlesworth D, Charlesworth B. 1987. Inbreeding depression and its evolutionary consequences. Annual Review of Ecology and cerns about the amount of failed effort contributed to Systematics 18:237–268. outcrossed progeny. Cheplick GP, Quinn JA. 1982. Amphicarpum purshii and the “pessimistic strategy” in amphicarpic annuals with subterranean fruit. Oecologia 52:327–332. Acknowledgements Clay K. 1983. The differential establishment of seedlings from chas- We would like to thank G. Clarke, S. Smith, S. McAllister mogamous and cleistogamous flowers in natural populations of and several cohorts of Plant Ecology Program interns the grass Danthonia spicata (L.) Beauv. Oecologia 57:183–188. for field assistance (complete list of interns available Coile NC, Garland MA. 2003. Notes on Florida’s endangered and threatened plants, 4th edn. Gainesville, FL: Florida Department at http://www.archbold-station.org/abs/staff/menges/ of Agriculture and Consumer Services/Division of Plant Industry. esmcvasst.htm). We thank Florida’s Endangered Plant Culley TM. 2002. Reproductive biology and delayed selfing in Viola Advisory Council, D. Bender and others at the US Fish pubescens (Violaceae), an understory herb with chasmoga- and Wildlife Service for their support and advice. Finally, mous and cleistogamous flowers. International Journal of Plant we thank Associate Editor Joseph Williams and review- Sciences 163:113–122. ers for their helpful comments and feedback. Culley TM, Klooster MR. 2007. The cleistogamous breeding system: a review of its frequency, evolution, and ecology in angiosperms. The Botanical Review 73:1–30. Sources of Funding Diaz A, MacNair MR. 1998. The effect of plant size on the expression of cleistogamy in Mimulus nasutus. Functional Ecology 12:92–98. The Florida Division of Plant Industry and the National Goodwillie C, Kalisz S, Eckert CG. 2005. The evolutionary enigma of Science Foundation. No grant numbers are provided mixed mating systems in plants: occurrence, theoretical expla- from the Division of Plant Industry since these were nations, and empirical evidence. Annual Review of Ecology, annual grants and the multiple numbers would take Evolution, and Systematics 36:47–79. lots of space. They can be provided if needed. The award IBM Corp. 2013. IBM SPSS Statistics for Windows, Version 22.0. number for the grant from NSF is provided but is not Armonk, NY: IBM Corp. highlights. James CW. 1957. Notes on the cleistogamous species of Polygala in southeastern United States. Rhodora 59:51–56. Jasieniuk M, Lechowicz MJ. 1987. Spatial and temporal variation in Contributions by the Authors chasmogamy and cleistogamy in Oxalis montana (Oxalidaceae). American Journal of Botany 74:1672–1680. S.M.K. completed final analyses for publication, wrote the Jones NT, Husband BC, MacDougall AS. 2015. Reproductive system manuscript and completed all edits. C.W.W. designed of a mixed-mating plant responds to climate perturbation by and implemented the initial project, collected data for increased selfing. Proceedings of the Royal Society B 280:20131336. most years and did preliminary analyses. S.J.H.C. col- Lande R, Schemske DW. 1985. The evolution of self-fertilization and lected data in the final years of the project and did many inbreeding depression in plants. I. Genetic models. Evolution preliminary analyses. E.S.M. assisted with writing and 39:24–40. Le Corff J. 1993. Effects of light and nutrient availability on chasmog- editing the manuscript for publication. amy and cleistogamy in an understory tropical herb, Calathea micans (Marantaceae). American Journal of Botany 80:1392–1399. Conflicts of Interest Lloyd DG. 1979. Some reproductive factors affecting the selection of self-fertilization in plants. American Naturalist 113:67–79. None declared. Lord EM. 1981. Cleistogamy: a tool for the study of floral mor- phogenesis, function and evolution. The Botanical Review 47:421–449. Literature Cited Lu Y. 2002. Why is cleistogamy a selected reproductive strategy in Albert LP, Campbell LG, Whitney KD. 2011. Beyond simple reproduc- Impatiens capensis (Balsaminaceae)? Biological Journal of the tive assurance: cleistogamy allows adaptive plastic responses Linnean Society 75:543–553. to pollen limitation. International Journal of Plant Science Masuda M, Yahara T. 1994. Reproductive ecology of a cleistoga- 172:862–869. mous annual, Impatiens noli-tangere L., occurring under differ- Bell TJ, Quinn JA. 1987. Effects of soil moisture and light intensity ent environmental conditions. Ecological Research 9:67–75. on the chasmogamous and cleistogamous components of Menges ES. 1999. Ecology and conservation of Florida scrub. In: reproductive effort on Dichanthelium clandestinum populations. Anderson RC, Fralish JS, Baskin JM, eds. Savannas, barrens, and Canadian Journal of Botany 65:2243–2249. rock outcrop plant communities of North America. Cambridge, Berg H, Redbo-Torstensson P. 1998. Cleistogamy as a bet-hedging UK: Cambridge University Press, 7–22. strategy in Oxalis acetosella, a perennial her. Journal of Ecology Mitchell-Olds T, Waller DM. 1985. Relative performance of selfed and 86:491–500. outcrossed progeny in Impatiens capensis. Evolution 39:533–544.

8 AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018 Koontz et al. – Patterns of chasmogamy and cleistogamy in Polygala lewtonii

Munguía-Rosas MA, Arias LM, Jurado-Dzib SG, Mezeta-Cob CR, Parra- Sun M. 1999. Cleistogamy in Scutellaria indica (Labiatae): effective Tabla V. 2015. Effects of herbivores and pollinators on fruit yield mating system and population genetic structure. Molecular and survival in a cleistogamous herb. Plant Ecology 216:517–525. Ecology 8:1285–1295. Myers RL 1990. Scrub and high pine. In: Myers RL, Ewel JJ, eds. Swift JA, Smith SA, Menges ES, Bassüner B, Edwards CE. 2016. Ecosystems of Florida. Orlando, FL: University of Central Florida Analysis of mating system and genetic structure in the endan- Press, 150–193. gered, amphicarpic plant, Lewton’s polygala (Polygala lewtonii). Oakley CG, Moriuchi KS, Winn AA. 2007. The maintenance of Conservation Genetics 17:1269–1284. outcrossing in predominantly selfing species: ideas and evi- US Fish and Wildlife Service. 1999. Multi-species recovery plan for dence from cleistogamous species. Annual Review of Ecology, the threatened and endangered species of South Florida. Vero Evolution, and Systematics 38:437–457. Beach, FL: US Fish and Wildlife Service. Schemske DW. 1978. Evolution of reproductive characteristics in Waller DM. 1979. The relative costs of self- and cross-fertilized Impatiens (Balsaminaceae): the significance of cleistogamy and seeds in Impatiens capensis (Balsaminaceae). American Journal chasmogamy. Ecology 59:596–613. of Botany 66:313–320. Schmitt J, Ehrhardt D, Swartz D. 1985. Differential dispersal of self- Waller DM. 1980. Environmental determinants of outcrossing in fertilized and outcrossed progeny in Jewelweed (Impatiens cap- Impatiens capensis (Balsaminaceae). Evolution 34:747–761. ensis). American Naturalist 126:570–575. Waller DM. 1984. Differences in fitness between seedlings derived Schoen DJ, Brown AHD. 1991. Whole- and part-flower self-pollina- from cleistogamous and chasmogamous flowers in Impatiens tion in Glycine clandestina and G. argyrea and the evolution of capensis. Evolution 38:427–440. autogamy. Evolution 45:1651–1664. Weekley CW, Brothers A. 2006. Failure of reproductive assurance in Schoen DJ, Lloyd DG. 1984. The selection of cleistogamy and het- the chasmogamous flowers of Polygala lewtonii (Polygalaceae), eromorphic diaspores. Biological Journal of the Linnean Society an endangered sandhill herb. American Journal of Botany 23:303–322. 93:245–253. Shaw CH. 1901. The comparative structure of the flowers inPolygala Weekley CW, Menges ES. 2012. Burning creates contrasting demo- polygama and P. pauciflora with a review of cleistogamy. graphic patterns in Polygala lewtonii (Polygalaceae): a cradle-to- Contributions from the Botanical Laboratory of the University of grave analysis of multiple cohorts in a perennial herb. Australian Pennsylvania 2:122–149. Journal of Botany 60:347–357. Steets JA, Ashman TL. 2004. Herbivory alters the expression of a Winn AA, Moriuchi KS. 2009. The maintenance of mixed mating by mixed-mating system. American Journal of Botany 91:1046–1051. cleistogamy in the perennial violet Viola septemloba (Violaceae). Stone JL. 2002. Molecular mechanisms underlying the breakdown American Journal of Botany 96:2074–2079. of gametophytic self-incompatibility. The Quarterly Review of Zeide B. 1978. Reproductive behavior of plants in time. The American Biology 77:17–32. Naturalist 112:636–639.

AoB PLANTS https://academic.oup.com/aobpla © The Authors 2017 9 Downloaded from https://academic.oup.com/aobpla/article-abstract/9/6/plx059/4600026 by guest on 15 February 2018